Alcohol and Other Drug Treatment for Parents and Welfare Recepients: Outcomes, Costs and Benefits. 2. Data Source: About Caldata


The primary source of data for this study is the California Drug and Alcohol Treatment Assessment (CALDATA). CALDATA was a pioneering large-scale study of the effectiveness, costs, and benefits of alcohol and drug treatment in California, using state databases, provider records, and follow-up interviews with clients in treatment. CALDATA was designed to reveal baseline, in-treatment, and post-treatment characteristics of a representative probability sample of 1991-92 clients in the four major types of treatment available to California residents who are eligible for public support such as MediCal (California's Medicaid program), state/county alcohol and drug treatment funding, or public disability insurance. The study's primary source of information was a voluntary survey of publicly supported clients.

A particular advantage to this data is that CALDATA was the first and is still the only available follow-up interview study to use random probability sampling of treatment populations rather than to select specific programs of interest or convenience. In other words, we selected with the objective that every individual in the treatment population would have a calculable and to the extent feasible equal chance of being selected. The main limitations on equality of chances were (a) our need to get sufficient sample numbers of each main program type so that we could study each program type in itself with reasonable precision, and (b) vagaries in program record-keeping or similar matters that made some clients easier to find and interview than others. We adjusted (weighted) all of the results reported here to reflect these variations in sampling probabilities, so that the results could be projected accurately back to the treatment population as a whole.

More than 36 percent of approximately 157,000 individuals represented by CALDATA we will call the large, represented group the "CALDATA treatment population," in contrast to the much smaller treatment sample who were actually interviewed reported having children in their household in the year prior to treatment. About 27 percent of the overall treatment population received AFDC or other welfare income before, during, or after treatment. About 38 percent of the treatment population were women. Additional details about the sample are discussed in Chapter 3.

The methods used in the CALDATA study are described in the original study publication (Gerstein et al., 1994) and in greater detail in the methodological report (Suter et al., 1994). A brief review is included here in order to help clarify the definitions, value, and limits of the information used.

CALDATA gathered information on five types of drug and alcohol treatment in California. The treatment types were:

  • Residential Treatment (21 providers selected)
  • Social Model Recovery Houses (23 providers selected)
  • Nonmethadone Outpatient (29 providers selected)
  • Methadone Programs two subtypes:
    • Methadone Maintenance Outpatient (18 providers selected)
    • Detoxification (19 providers selected)

The two methadone provider groups were selected separately, but the samples in fact overlapped since most methadone providers offered both detoxification and maintenance treatment using the same facility and staffing.

These treatment types differ characteristically from each other in a variety of ways. A brief description of each follows:

  • Residential Treatment in general. A variety of recovery service approaches are employed in residential settings, which can provide heavily structured and controlled environments. Some residential programs are oriented more towards individual counseling and a classical staff/therapist model; others stress group interaction or a gradual climb through successive roles and responsibilities as a milieu for assimilating new ideas, norms, and behaviors.
  • Social Model Recovery Houses. Seen more in California than other states, these are a particular type of residential program that which focus on recovering alcoholics, stressing peer support and communal sober living.
  • Nonmethadone Outpatient. Outpatient programs, exclusive of those providing daily methadone doses, encompass great variety, from one hour/week one-to-one counseling that may be focused on practical, emotional, spiritual, or other issues; to daily or multiple weekly individual or group sessions that may focus on these matters or on the 12 Steps (as in Alcoholics Anonymous or Narcotics Anonymous). Some programs include substantial medical or psychiatric elements, others none at all.
  • Methadone Maintenance. In maintenance, a stable daily oral dose of methadone hydrochloride, accompanied by other available nonresidential services such as counseling, is provided to formerly heroin-dependent clients on a long-term basis. Maintenance is open only to those who have either relapsed to heroin use following two or more previous treatments or are pregnant. Methadone in appropriate doses prevents withdrawal symptoms and maintains a level baseline of physical comfort and functioning with virtually no psychological or physiological impairment.
  • Methadone Detoxification. Methadone detoxification means support for planned with-drawal from heroin (or sometimes other opiate) dependence using a gradually tapering dose of methadone hydrochloride, lasting a maximum of 21 days.

Of the total treatment population, 19 percent were in the two types of residential programs, which are combined for the purposes of this report; 35 percent were in nonmethadone outpatient programs, 39 percent had been discharged from methadone programs, and almost 7 percent were being treated in continuing methadone programs. Women were more likely to be in non-methadone outpatient programs than men (39% versus 32%), and men were more likely to be in residential programs (21% versus 17%). Clients in residential programs of either gender tended to stay for a shorter period if they had children in their household prior to treatment (See Figure 2.1). For the other types of treatment, however, the presence of children in the household was not related to the length of stay in treatment.

Figure 2.1 Length of Stay by Gender, Children in
Household, and Type of Treatmen


Clients were selected at random from discharge (or in-treatment) lists developed on site at cooperating providers. Sixteen counties, 97 providers, and approximately 3,000 clients who were in treatment or were discharged between October 1, 1991 and September 30, 1992 were selected into the study sample. As authorized by federal and state law and permitted by consent obtained routinely on admission to treatment, the program records of clients selected for the follow-up sample were read and abstracted to determine additional important research information and to verify the self-reported data. Using a combination of methods including letters, postcards, telephone calls, visits to last known addresses, contacting relatives or institutional connections, and searching various accessible public records, CALDATA staff sought to locate members of the sample and seek their participation in the study. Respondents received a cash honorarium of $15 for completing an interview. In order to protect the privacy of respondents, strict confidentiality was maintained throughout the data collection period. The methods used to protect confidentiality were approved by the California Health and Welfare Protection of Human Subjects Committee.

Out of the 3,000 clients sampled, 1,858 were successfully contacted and interviewed during a 9-month field period.5 These clients were drawn from 83 cooperating providers (out of 97 sampled; we note that 9 of 14 noncooperating providers were part of two large chains of private, proprietary methadone providers) in 15 counties (out of 16 sampled from 58 counties in the state, many of which were sparsely populated mountainous counties whose residents would travel to more populated areas for treatment). The client follow-up interview was developed for CALDATA based on extensive work with previous research studies. The questionnaire took approximately one hour and fifteen minutes to administer on average. Follow-up interviews occurred an average of 15 months after treatment, with the longest interval being 24 months. All time-sensitive questionnaire results were adjusted to control for the length of the after-treatment period covered by each interview. Part of the sample was comprised of individuals who were in continuing methadone maintenance treatment, since this type of treatment is typically longer term than other services.

Program records data on the respondent sample was compared with program records data on the nonrespondents to determine whether the sample had drifted to any extent away from the population to be represented. There were virtually no statistically significant differences between the responding and nonresponding clients on scores of program-level variables. The key comparisons, including all significant differences, are summarized in Table 2.1. We note that the sample overrepresents women and Hispanics, which is a pattern of higher contact and cooperation rates that is common to most surveys in the United States. The survey slightly underrepresents employed individuals and overrepresents those whose treatment was primarily paid for by public sources. In general, we believe these results reflect the relative ease of locating and interviewing individuals who can be found at home rather than at work and for whom the cash honorarium would be a greater incentive. We also found that, notwithstanding the firm guarantee of confidentiality, employed individuals were more reluctant to participate in the study once they discerned that the subject matter might be viewed very negatively by an employer, particularly in light of the deteriorating economic situation in California at the time interviews were taking place (see Chapter 5).

Characteristics Interviewed
(Base N
Not Interviewed
(Base N
Table 2.1 Comparison of Sample Interviews and Noninterviewed Cases in CALDATA Using Data from Administrative Records of Cooperating Providers
Sample averages (means)    
Length of sample episode (months) 2.8 (1570) 2.7 (1108)
Age at admission 33.3 (1523) 33.5 (1068)
Education (1=did not complete high school, 2=HS grad or CED, 3=Beyond HS)* 1.8 (1531) 1.9 (1090)
# Treatment services recieved 2.9 (1025) 2.8 (733)
# Medications prescribed during treatment 1.8 (1580) 1.9 (1114)
% with physician notes at admission 48% (1585) 50% (1116)
% with physician notes discharge 13% (1580) 12% (1115)
% with physician notes at any other time 29% (1576) 29% (1114)
% with planned treatment > 25 days** 34% (1821) 35% (1183)
% with self as primary referral source 46% (1410) 46% (1015)
% with leagal system as primary referral 22% (1410) 23% (1015)
% with public as primary payment source** 50% (1316) 45% (871)
% female** 38% (1585) 33% (1113)
% Black (African American) 15% (1578) 15% (1115)
% Native American 1.5% (1578) 1.1% (1115)
% White 76% (1578) 78% (1115)
% Hispanic or Latino** 37% (1319) 30% (929)
% who ever used needles to inject drugs** 72% (1060) 71% (707)
% with cocaine as primary drug at admission** 42% (1471) 17% (1046)
% with heroin as primary drug at admission** 42% (1471) 40% (1046)
% with alcohol as primary drug at admission** 27% (1471) 29% (1046)
% with psychiatric history at admission 12% (803) 12% (609)
% employed at admission** 21% (1515) 27% (1068)
% with chronic med. condition at admission** 35% (923) 31% (700)
% with length of treatment > 25 days 58% (1576) 58% (1107)
% tested for drug or alcohol abuse during sample episode** 65% (1066) 64% (759)
% completing treatment plan before discharge** 32% (1812) 31% (1180)
% with aftercare plan stated in record 35% (1821) 35% (1180)

Base n's in parentheses are the numbers of interviewees who had nonmissing data (in the program records) for the item. The maxima are 1,812 of those interviewed and 1,180 of those not interviewed. For some cases, so many items were missing from or inconsistent in their records that these cases were omitted from this comparison.

*Significant difference between respondents and nonrespondents based on two sample t test, two tail, alpha=.05.

**Significant difference between respondents and nonrespondents based on chi-square test of independence, alpha=.05.